Engage Coaching for Caregivers: a pilot trial to reduce loneliness in dementia caregivers

Abstract

INTRODUCTION:

There are no evidence-based interventions for reducing loneliness in family caregivers of people with dementia (ADRD caregivers), despite heightened risk. We examined feasibility, acceptability, and potential efficacy of a brief behavioral intervention, Engage Coaching for Caregivers, to reduce loneliness and increase social connection for older ADRD caregivers experiencing stress and loneliness.

METHODS:

A single-arm clinical trial of 8 individual sessions of Engage Coaching delivered remotely. Outcomes assessed 3-months post-intervention included loneliness and relationship satisfaction (co-primary) and perceived social isolation (secondary).

RESULTS:

Engage Coaching was feasible to deliver, with n=25 of 30 enrolled completing at least 80% of sessions. 83% indicated the program met expectations and 100% reported the program was suitable and convenient. Improvements were observed in loneliness (standardized response mean [SRM] =0.63), relationship satisfaction (SRM=0.56), and perceived social isolation (SRM=0.70).

DISCUSSION:

Engage Coaching is a promising behavioral intervention to enhance social connection for older ADRD caregivers.

1. INTRODUCTION

Providing care for a family member with Alzheimer’s Disease and related dementias (ADRD; hereafter ADRD caregivers) increases risk for poor mental and physical health, and reduced quality of life (National Academies of Sciences, 2021). One pathway whereby dementia caregiving may reduce health and quality of life is by reducing social connection (Kovaleva et al., 2018; Van Orden & Heffner, 2022)—the quantity and quality of social ties that individuals have with other people (Holt-Lunstad et al., 2017). Dimensions of social connection that may be impacted by caregiving include social isolation (structural dimensions of social networks), social support (functional aspects of social networks), and subjective perceptions of social networks (e.g., satisfaction with relationships, loneliness, belonging; Van Orden & Heffner, 2022). Providing care for a loved one with dementia can directly impact these dimensions of social connection by making it challenging for caregivers to leave the home (Robison et al., 2009), to receive social support (Vasileiou et al., 2017), and to prioritize their own needs for social interaction (Clark & Bond, 2000). Dementia can also impact the quality of the relationship between the caregiver and the person with dementia, including reductions in emotional intimacy (Beeson, 2003; Vasileiou et al., 2017). Social connection may be especially salient in promoting the well-being of ADRD caregivers given that the presence of social support may buffer against the negative effects of stress produced by caregiving demands (Leggett et al., 2021; Wang et al., 2018). Despite the potential of social connection to buffer caregiving stress and promote well-being, there are no evidence-based interventions for promoting social connection in older ADRD caregivers (Van Orden & Heffner, 2022).

This study examines the feasibility, acceptability, and potential efficacy of a brief behavioral intervention, Engage Coaching for Caregivers (hereafter Engage Coaching), to reduce loneliness and increase social connection for ADRD caregivers experiencing stress and loneliness. While the study was not designed in response to the COVID-19 pandemic, all subjects were enrolled while social distancing protocols were in place; thus, all subjects addressed loneliness in the context of ADRD caregiving during the COVID-19 pandemic. The intervention itself was unchanged because Engage Coaching addresses challenges in leaving the home as a barrier to social connection, given this is a challenge commonly faced by ADRD caregivers.

Engage Coaching has three primary components: identifying contributors to loneliness within the context of caregiving as well as valued domains of social connection to serve as personalized intervention targets; creating action plans to increase social connection; addressing barriers to social connection. The goals of Engage Coaching are bolstering motivation for improving social connection, teaching an intentional and active approach to social interactions and relationships, developing problem solving skills, and engaging in repeated behavioral practice with social behaviors that promote feelings of positive social connection. The primary principle underlying Engage Coaching is ‘social reward exposure,’ meaning that as participants engage in social activities that produce positive, valued outcomes (e.g., emotional support, shared experiences), this motivates participants to continue such activities. Our conceptual model (Figure 1) proposes that increased social reward should impact the target mechanism of social connection, operationalized as loneliness and satisfaction with social relationships and activities, which should lead to reduced caregiving strain and improved quality of life. While conceptualized as mechanisms whereby Engage Coaching impacts health outcomes, our pilot trial was grounded in an experimental therapeutics approach (Insel & Gogtay, 2014) and thus focused primarily on whether the intervention could impact the proposed mechanisms of action. Thus, the co-primary outcomes for this trial are loneliness and satisfaction with social relationships and activities.

Figure 1. Conceptual model for Engage Coaching for Caregivers.

Note: Dashed line indicates focus of the current Stage I clinical trial to test engagement of target mechanism.

We conducted a Stage 1 (NIH Stage Model; Onken, 2022; Onken et al., 2014) intervention development study using a single-arm clinical trial of Engage Coaching, with 8 individual coaching sessions provided remotely over no more than 3 months, with primary outcomes assessed post-intervention (3-month follow-up). Our first aim examined feasibility of study procedures (focused on retention) and acceptability of the intervention for ADRD caregivers. We anticipated at least 80% of caregivers enrolled would complete the three-month follow-up interview (based on our previous study of Engage with lonely older adults) and that at least 75% of participants would complete 5 or more sessions.

Our second aim was to examine a signal for potential efficacy on co-primary outcomes of loneliness and satisfaction with social roles and activities, as well as a secondary outcome of perceived social isolation (an alternate measure for loneliness given that there are no established measures for clinical trials for loneliness). Our third (exploratory) aim examined supplemental mechanisms whereby Engage Coaching may reduce loneliness—increasing emotional or instrumental support—as well as potential health outcomes—quality of life, caregiving strain, depression, and anxiety symptoms.

2. METHODS

2.1. Participants

Participants were 30 adults aged 50 or older who: 1) identified as providing care for a loved one (family member or other close tie) with ADRD per self-report (“Do you provide care or support for a family member [or non-family loved one] diagnosed with any type of dementia?”); 2) endorsed clinically significant caregiving stress (score of greater than 11 on Perceived Stress Scale (Cohen et al., 1983) and/or a score of 5 or greater on the Modified Caregiver Strain Index (Thornton & Travis, 2003); and 3) reported clinically-significant loneliness on the UCLA 3-item Loneliness Scale (score of 6 or greater) (Hughes et al., 2004). Exclusion criteria were inability to communicate in English (primary Spanish speakers were invited to participate in a companion study on adapting Engage Coaching for Hispanic/Latino caregivers), recent or current psychosis, significant cognitive impairment on the Telephone Interview for Cognitive Status (Knopman et al., 2010), hearing problems that precluded completion of the intervention, and inability to complete coaching sessions via video or phone call. Level of depressive symptoms and ongoing psychiatric treatment were not inclusion/exclusion criteria but were assessed at baseline and follow-up. Type of dementia diagnosis in the person being cared for and frequency of care provided were not inclusion/exclusion criteria, but were assessed via self-report (“What is your loved one’s diagnosis? and “how often do you provide care for your loved one?”).

2.2. Procedures

The study’s timeline with regards to the COVID-19 pandemic is important to consider given the significant impact of social distancing protocols on social contact. This study was designed and planned prior to the COVID-19 pandemic and was slated to start enrolling in February 2020, but recruitment efforts were paused and all study procedures—both assessments and interventions—were adapted for remote delivery to comply with social distancing precautions. Subjects were enrolled between July 2020 and April 2021, with no change in study procedures during this time. Participants were recruited from ResearchMatch (Harris et al., 2012)—a web-based recruitment registry (n=19)—and the Healthy Aging Research Program research registry at the University of Rochester Medical Center (n=11). Potential subjects from both recruitment sources were invited to complete eligibility online self-report surveys that assessed loneliness and caregiver status (see section 2.1). Respondents who scored a 6 or greater on the 3-item UCLA Loneliness Scale and responded “yes” to the question about caregiving were invited to complete baseline interviews with trained study assessors to confirm eligibility and characterize the sample. Respondents who identified as Hispanic/Latino were invited to participate in a companion study adapting Engage Coaching for the unique needs of this sub-population of caregivers. All participants were offered the study intervention and were invited to complete follow-up interviews at 3-months post-intervention. This study was approved by the Institutional Review Board of the University of Rochester and participants provided written informed consent. The study was registered on clinicaltrials.gov (NCT04176601). Data were collected and managed in REDCap (Harris et al., 2009).

2.3. Measures

We measured feasibility with the number enrolled and retained at follow-up as well as number of coaching sessions completed. Acceptability was measured with the Treatment Acceptability and Preferences (TAP) Scale (Sidani et al., 2009) which includes four items (rated on a 5-point scale—not at all [0] to very much [4]) that assess dimensions of acceptability—appropriateness of the intervention, suitability (feasibility), effectiveness, and convenience. Items were created to assess subjects’ preferences for: video calls versus in person, number of sessions, and addition of booster sessions.

2.3.1. Primary and secondary outcomes:

Loneliness (primary outcome) was measured with the UCLA Loneliness Scale Version 3 (Russell, 1996) that has been validated with older adult samples (Ausin et al., 2019). The scale has 20 items (e.g., "How often do you feel alone?"), rated on a four-point scale ("never" 1, "rarely" 2, "sometimes" 3, or "often" 4). Total scores range from 20 to 80, with higher scores representing greater loneliness and scores of 43 and above representing clinically meaningful loneliness in older adult samples (Lee et al., 2019).

Satisfaction with social relationships and activities (co-primary outcome) was measured with the Satisfaction with Social Roles and Activities computerized adaptive test (CAT) of the Patient Reported Outcomes Measurement System (PROMIS) (Hahn et al., 2016; Hahn et al., 2010). It produces T scores with a mean of 50 and standard deviation of 10; higher scores indicate greater satisfaction. Perceived social isolation (secondary outcome) was measured with the PROMIS Perceived Social Isolation CAT (Hahn et al., 2010; Karayannis et al., 2019) with T scores with a mean of 50 and standard deviation of 10. Higher scores indicate greater perceived isolation.

2.3.2. Exploratory outcomes:

Caregiving strain was measured with the Modified Caregiver Strain Index (MCSI), which has 13 items (scored 0-2) that measure perceived negative impacts of caregiving, including several domains of strain—financial, physical, psychological, social, and personal (Thornton & Travis, 2003). Scores range from 0 to 26, with higher scores indicating greater caregiver strain. Depressive and anxiety symptoms were measured with PROMIS CATs (Cella et al., 2016; Schalet et al., 2016) with higher T-scores (mean of 50, standard deviation of 10) indicating greater symptomatology. Quality of life was measured with the total score from the World Health Organization Quality of Life Scale (WHOQOL-Bref) (Skevington et al., 2004) which has a population mean of 50 and higher scores indicating greater than average quality of life.

2.4. Intervention

Engage Coaching for Caregivers is an adaptation of Engage Psychotherapy, an evidence-based psychotherapy for depression in later life (Alexopoulos et al., 2020) that has also been used to address social disconnection for older adults reporting loneliness (Van Orden et al., 2021) with positive effects on depression and social-emotional quality of life. Given high acceptability and flexibility to tailor the intervention to individual pathways to social disconnection, we adapted Engage Psychotherapy to be responsive to the specific needs and contexts of ADRD caregivers experiencing loneliness. The primary components of Engage Psychotherapy were unchanged for this trial, including that it is a stepped, modular intervention that addresses barriers to action plan implementation and challenges in processing positive outcomes when exposed to rewarding activities. When barriers are identified, simple behavioral interventions to address barriers are added to action plans. Modifications for this trial involved tailoring the rationale to address social connection in the context of caregiving; tailoring psychoeducational materials to address how caregiving can impact social connection; and adding barrier strategies to address common barriers to social connection in caregivers (discussed below).

Engage Coaching helps caregivers increase their awareness of the importance of social connection as well as motivation to address it; teaches problem solving skills to address barriers to social engagement (due to caregiving roles); and provides behavioral practice with social engagement. In addition to addressing barriers specified in Engage Psychotherapy when needed (i.e., negativity bias, apathy, emotion dysregulation), additional barrier strategies were used for managing emotional reactions around being a caregiver (e.g., guilt, anxiety), practical barriers produced by caregiving (e.g., difficulty leaving the home), and lack of knowledge/skills about dementia caregiving. Eight brief sessions (typically 30 minutes) were provided weekly over no more than three months (via video call or phone). Table 1 lists the content of each session. The main therapeutic activity is ‘Action Planning,’ in which caregivers develop a concrete plan to engage in a socially rewarding activity before their next coaching session; the plan addresses barriers to completing the plan as needed (e.g., need for someone to stay with the person with dementia; guilt about taking time for oneself; low energy). Caregivers selected goals each week based on aspects of social connection they value most and that were disrupted by caregiving. For example, “My goal is to be more open with my kids about how I’m feeling emotionally about caregiving.” Engage Coaches were doctoral level clinical psychologists (n=3) and graduate students in clinical psychology (n=2) who were not involved in study assessments. Training involved review and discussion of the intervention manual with the study’s lead author, role plays, and watching session recordings. Weekly clinical supervision was provided by the study’s lead author who is a certified Engage trainer; supervision included review and discussion of session recordings to ensure fidelity to the intervention. The manual will be available online to download at the website for the Rochester Roybal Center for Social Ties & Aging Research (https://research.son.rochester.edu/rocstarcenter/).

Table 1.

Engage Coaching for Caregivers Session Content

Session	Coach Task 1	Coach Task 2	Coach Task 3
1	Orient to Engage: structure of therapy, session frequency, rationale.	Gather info: Current social network, relationship with person caring for, typical social contact—now and before caregiving; exploration of values with ‘social pie of life.’	Begin first action plan: Brainstorm Social Activity List; teach action plan process with a relatively easy goal
2	Set agenda & mood/stress check	Review homework; begin discussion of barriers to social engagement via any barriers experienced completing homework; create barrier list	Create action plan & review expectation for homework
3	Set agenda & mood/stress check-in	Review homework; introduce Barrier Strategy 1 (if needed)	Create action plan; discuss doing an extra action plan on own (or use time for integrating barrier strategy)
4	Set agenda & mood/stress check-in	Review homework	Create action plan
5	Set agenda & mood/stress check-in	Review homework; add Barrier Strategy 2 (if needed)	Create action plan
6	Set agenda & mood/stress check-in	Review homework; begin discussion of termination	Create action plan
7	Set agenda & mood/stress check-in	Review homework; continue discussion of termination	Create action plan; homework to consider changes made & ways to maintain gains
8	Set agenda & mood/stress check-in	Wrapping up: Accomplishments regarding “valued life”; areas to continue to work on, concrete steps to take going forward	Saying goodbye

2.5. Data Analytic Strategy

To examine the mechanism of missing data, participants who were lost to follow-up or withdrew (n=5) were compared on several baseline variables, with no differences on age, sex, marital status, living with care receiver, relationship with the care receiver, outcome variables (loneliness, social functioning, perceived isolation), history of a mental health diagnosis, caregiver stress, depression severity, nor anxiety severity. Examining demographic characteristics of those who were lost to follow-up indicated that two out of five were non-White (1 Black, 1 multiracial) and four out of five were working or looking for work.

Aim 1 examines enrollment, retention, and adherence numbers. For Aim 2, the effect size we report is the standardized response mean (SRM), which assesses clinically meaningful change. The SRM represents the ratio of the mean change to the standard deviation of that change and is a form of Cohen’s effect size index useful for indicating responsivity to an intervention given that it takes into account variability in change over time and is more conservative than scores using the standard deviation of the baseline score (Revicki et al., 2006). A SRM of .3 is considered the lower bound for clinically meaningful change in health outcomes (Revicki et al., 2006). We did not conduct a formal power analysis given that this was a pilot trial; however, we selected a target sample size of 30 participants given anticipated adherence, attrition, and effect sizes from our prior trial of Social Engage with n=32.

3. RESULTS

Table 1 characterizes the sample and Table 2 includes baseline and follow-up scores for primary, secondary, and exploratory outcomes. The average age was 61 (std 8.51); most were female (80%), identified as White (92%), and were married (57%). Participants reported providing care for a parent (53%), spouse (33%), or another close tie (family member, friend, romantic partner, 20%) and most reporting living with the person with dementia (83%) and providing daily or full-time care (80%). The most common type of dementia was Alzheimer’s Disease (33%). Average depression and anxiety scores were in the mild range. Scores for dimensions of social support—emotional, and instrumental—fell in the normal range, whereas loneliness was elevated and satisfaction with social relationships was low. Scores for general stress and caregiver strain were elevated. These characteristics are consistent with our objective of enrolling a sample of ADRD caregivers appropriate for a behavioral intervention to reduce loneliness.

Table 2:

Participant characteristics at baseline

	Mean (sd) or N (%)
Age	61.37 (8.51)
Female	n=27 (80%)
Race (White)1	27 (92%)
Lives with PWD	25 (83%)
Married2	n=17 (57%)
Relationship to PWD
Parent	n=16 (53%)
Spouse	n=10 (33%)
Other3	n=6 (20%)
Type of dementia
Alzheimer’s Disease	n=10 (33.3%)
Lewy Body/Parkinson’s	n=4 (13.3%)
Vascular	n=3 (10%)
Fronto-temporal	n=1 (3%)
Mixed	n=2 (6.7%)
Not specified	n=10 (33.3%)
Provides daily or full-time care	n=24 (80%)
General stress level (baseline)4	18.37 (6.16)

Notes:

¹Other race was n=2 black and n=1 multiracial

²Marital status was 9=single never married and n=4 divorced

³Other relationships to PWD were n=1 cousin, n=1 friend, n=1 uncle, and n=1 romantic partner (never married)

⁴Perceived Stress Scale.

Abbreviations: PWD, person with dementia.

Aim 1 examined feasibility and acceptability of Engage Coaching. A CONSORT diagram (Eldridge et al., 2016) depicts enrollment and retention outcomes (Figure 2). Out of 117 potential participants screened over 10 months, 81 (54%) did not meet inclusion criteria for loneliness and 18 (15%) were not interested in the study, while 36 (31%) were assessed for eligibility (see Figure 2). A total of 30 caregivers enrolled over 10 months, 25 completed 3-month follow-up assessments (83%), and 25 completed at least 5 coaching sessions (83%), meeting or exceeding targets for feasibility. The majority (70%, n=21) completed all 8 coaching sessions, 6 subjects completed between 4 and 7 sessions (20%), 2 subjects completed only 1 session (10%), and only 1 subject did not complete any sessions. Subjects who provided reasons for discontinuing sessions (n=4) all stated that caregiving responsibilities made it too challenging for them to participate, including one care receiver who contracted COVID, another care receiver who died during the study, and another care receiver who moved into an assisted living facility. Results from the TAP Scale supported acceptability, with 100% reporting the program was appropriate (mean=3.90, std 0.30), suitable (mean=3.71, std 0.46), and convenient (mean=3.67, std 0.56) for individuals caring for a loved one with dementia, and 95% reporting the program is effective for individuals caring for a loved one with dementia (mean=3.71, std 0.56). To inform subsequent optimizations, we also asked about the structure of the intervention, with most subjects reporting that 8 sessions was adequate (62%), booster sessions would be useful (81%), and video call (versus in person) format was preferable (100%).

Figure 2. CONSORT Diagram.

Results for Aim 2 (Table 2) indicated clinically-significant improvement (i.e., average SRM of .3 or greater) for both co-primary outcomes and the secondary outcome, with 60% of subjects demonstrating an SRM of .3 or greater for loneliness (average standardized response mean [SRM] =0.63), 48% for satisfaction with social relationships and activities (average SRM=0.56), and 72% for perceived social isolation (average SRM=0.70). Results indicated improvement in several exploratory outcomes, with 52% of subjects demonstrating an SRM of .3 or greater for instrumental social support (average SRM=0.30), 71% for caregiving strain, and 80% for quality of life. Results for emotional support, depression, and anxiety did not meet our pre-specified criteria for meaningful change. Given this is a pilot study, our focus is on effect sizes; however, Table 3 also includes 95% confidence intervals for mean differences for all outcomes between baseline and 3-months, all of which indicate statistically significant change for outcomes with clinically-significant effects.

Table 3.

Effect sizes & number of responders at 3-month follow-up for primary, secondary, and exploratory outcomes

Outcome		Mean (std dev)		Effects at 3-month follow-up
		Baseline	3-months1	Mean diff,2 std error (95% CI)	SRM3	Responders4
Co-Primary	Loneliness (UCLA)	49.57 (9.87)	44.44 (10.01)	4.32, std 1.37 (1.49 – 7.15)	0.63	15 (60%)
Co-Primary	Satisfaction with relationships	39.99 (4.97)	44.92 (6.75)	4.70, std 1.67 (1.26 – 8.15)	0.56	12 (48%)
Secondary	Perceived isolation (PROMIS)	54.62 (5.95)	50.74 (6.10)	3.68, std 1.10 (1.59 – 6.14)	0.70	18 (72%)
Exploratory	Emotional support (PROMIS)	46.21 (7.86)	47.50 (6.29)	1.20, std 1.21 (1.31 – 3.70)	0.20	7 (28%)
Exploratory	Instrumental support (PROMIS)	45.62 (6.95)	47.37 (7.53)	1.07, std 0.71 (0.40 -– 2.54)	0.30	13 (52%)
Exploratory	Caregiving strain (MCSI)	16.60 (4.55)	11.88 (6.40)	8.34, std 1.67 (1.26 – 8.15)	0.83	17 (71%)
Exploratory	Quality of life (WHOQOL)	55.20 (9.59)	65.45 (13.46)	9.18, std 2.00 (5.04 – 13.32)	0.92	20 (80%)
Exploratory	Depression	55.30 (6.31)	53.56 (6.53)	1.37, std 1.05 (−.80 – 3.54)	0.26	12 (48%)
Exploratory	Anxiety	57.43 (6.92)	57.20 (4.51)	0.51, std 1.10 (−2.79 – 1.77)	0.09	9 (36%)

Notes:

¹At 3-months, 5 subjects did not complete assessments.

²Mean differences are calculated using baseline scores for those without missing data at 3-months, which differ slightly from baseline values depicted in this table.

³SRM is the standardized response mean, an effect size that represents the ratio of the mean change to the standard deviation of that change. Greater SRM values indicate greater improvement for the variable in question because change from baseline to follow-up was calculated as base – 3-month for loneliness, perceived isolation, strain, depression, anxiety, but 3-month – base for social functioning, emotional support, instrumental support, companionship to facilitate ease of interpretation.

⁴Responders are those subjects who demonstrated an SRM of at least 0.3.

4. DISCUSSION

Results indicate that Engage Coaching for Caregivers is feasible and acceptable in the context of a clinical trial for adults aged 50 and older who are providing care for a loved one with dementia and who are at risk for negative health outcomes due to elevated caregiving stress and loneliness. Results indicate that 8 individual coaching sessions was feasible for participants and that they perceived benefits. Most subjects demonstrated clinically meaningful improvement in loneliness—60% as assessed by the UCLA Loneliness Scale (primary outcome) and 72% as assessed by the PROMIS Social Isolation scale (secondary outcome). Prior research on Engage Psychotherapy for depression found a response rate of approximately 58% (Alexopoulos et al., 2020) suggesting that this adaptation produced a comparable response. This study was conducted during the COVID-19 pandemic, indicating that subjects’ experiences of loneliness were shaped not just by their caregiving experiences, but limitations in social contact necessitated by social distancing, which must be considered when interpreting findings.

Clinically meaningful improvement occurred for approximately half the sample for two additional dimensions of social connectedness—satisfaction with social relationships (48%) and instrumental social support (52%). Given that the specific targets of action planning in Engage Coaching are personalized to individual causes of loneliness, it is perhaps not surprising that not all participants improved on all dimensions of social connectedness that may cause loneliness: participants and coaches collaborated to identify whether addressing a lack of emotional or instrumental support or lack of social activities would be the most potent means to reduce loneliness. It could also be that these dimensions of social connectedness may be more challenging to improve for some caregivers, and/or that the COVID-19 pandemic may have made some dimensions of social connectedness more resistant to change due to limited services or opportunities for in-person interaction.

Several exploratory outcomes that may represent health outcomes impacted by Engage Coaching demonstrated meaningful change, with 71% demonstrating improvement in caregiving strain and 80% in quality of life. Results did not support depression or anxiety as potential outcomes (or mechanisms) whereby the intervention may have an impact. Given that the intervention was designed to target loneliness, positive effects on loneliness but lack of effect on depression and anxiety suggests that Engage Coaching is specifically targeting loneliness and not psychological distress more generally. In addition, the mechanism whereby Engage Coaching reduces loneliness appears to be improving dimensions of social connectedness (through increasing social reward) rather than general behavioral activation, which would be consistent with reductions in depressive symptoms.

While not initially designed for remote delivery, the COVID-19 pandemic necessitated this change for all participants, the majority of whom said that they preferred video or phone sessions due to challenges with scheduling or leaving their care receiver home alone. Remote interventions may reduce barriers to participation in interventions for many caregivers. Results should also be considered within this context. It could be that acceptability of Engage Coaching was enhanced given the relative salience of social isolation at a societal level. It could also be the case that social relationships would be more or less malleable in a different social context.

The results of this trial raise possibilities to test in future trials for how ADRD caregiving interventions may impact social relationships and quality of life. One possibility is the stress buffering effects of social connection: our results indicated reduced loneliness and increased relationship satisfaction, as well as declines in caregiving strain, which could be due to perceptions of caregiving being less burdensome (i.e., reduced strain) when caregivers feel meaningfully connected to others. Another possibility is that caregiving strain decreased because social connection directly impacted health behaviors such as sleep, nutrition, and physical activity, all of which are related to stress and quality of life (Holt-Lunstad, 2021). Future research is needed to clarify whether loneliness or caregiving strain changes first (temporally) and thus which variable should most aptly be labeled the mechanism versus outcome. Our trial specified loneliness as the primary outcome given our objective was to conduct an experimental therapeutics trial to examine whether Engage Coaching impacts the target mechanism of loneliness before proceeding to a fully powered trial that examines loneliness as a mediator of health outcomes, including caregiving strain and quality of life. Thus, these results are preliminary and do not provide a full test of our conceptual model. We plan to analyze the content of subjects’ action plans to gain a fuller, more nuanced understanding of how subjects used coaching sessions and what they perceived as beneficial. Doing so may indicate hypotheses to test regarding mechanisms and temporality (i.e., which outcomes change first).

Given that caregiving roles, skills, and responsibilities evolve over the course of the illness, future work is needed to examine durability of effects over longer-term follow-up. It will be important to know if subjects continue to prioritize social connection and are able to use new skills to stay connected in the face of new caregiving demands or changes that impact feelings of loneliness (e.g., death of the person with dementia). Most participants reported that they believed ‘booster sessions’ would help them maintain gains. Future work could examine whether booster sessions are needed for all subjects or whether some caregivers may benefit most, such as those with especially challenging caregiving situations or less knowledge and skills about dementia caregiving. For those caregivers, supplemental interventions that teach caregiving skills to promote mastery, such as the SAVVY Caregiver Program (Griffiths et al., 2018; Hepburn et al., 2003) could be useful to maintain gains in social connection. Caregivers with fewer financial and support resources may benefit from care management, in addition to Engage Coaching, so that social connection can be prioritized.

Findings should be considered in the context of limitations. First, given the non-randomized design, outcomes could primarily be due to the feelings of connection with Engage Coaches, suggesting improvements may not be maintained over the long-term. However, given that sessions were brief (typically 30 minutes) and that most participants created ‘action plans’ focused on social goals each week, it seems unlikely that all benefit was due to the emotional support of coaches. However, a randomized trial is needed to confirm findings. Second, we could not enroll individuals who were unwilling to complete remote sessions and who were not able to provide informed consent for the study remotely (i.e., necessitated an email address). While we encouraged subjects to use a web camera, this was not required and subjects could call into the Zoom meeting with a landline if they preferred. Thus, while subjects were not required to have a high level of technological sophistication or comfort, future research is needed to ensure that interventions provided remotely are accessible to all who could benefit. Third, Engage Coaches were pre-doctoral and postdoctoral level clinical psychologists, limiting generalizability. Future studies are needed to demonstrate that Engage Coaching can be delivered by a range of professionals, including those who work in community agencies that provide services to caregivers, such as Area Agencies on Aging, the Alzheimer’s Association, and primary care clinics. Given that Engage Psychotherapy was developed with dissemination and implementation considerations in mind, we anticipate that Engage Coaching for Caregivers will be similarly easy to deliver with fidelity by a diverse range of professionals. Fourth, despite recruiting nationwide, our sample was relatively homogenous with regards to race and ethnicity in part because potential subjects who identified as Hispanic/Latino were referred to our companion study on Engage Coaching for Hispanic/Latino caregivers and also because the COVID-19 pandemic limited our ability to recruit through in-person venues in our community that serve more racially diverse groups of older adults. Thus, future work is needed to confirm generalizability of our findings to more diverse caregivers and integrate results with our companion study on modifications needed for Engage Coaching for Hispanic/Latino caregivers. Although our sample is too small to speculate how race or ethnicity may have impacted treatment retention, it is well-established that contextual factors such as race, ethnicity, and culture impact caregiver needs (Pinquart & Sorensen, 2005) to create unique rewards and challenges supporting the utility of adaptations to optimize generalizability of treatment effects. It is also worth exploring in future work whether caregivers with fewer financial resources or those currently working full-time might have more difficulty prioritizing social connection and loneliness. Finally, as mentioned above, our study was conducted during the COVID-19 pandemic and results may not generalize to other contexts, necessitating replication.

Taken together, results support further study of Engage Coaching for Caregivers as a means to reduce loneliness in ADRD caregivers. Given promising findings regarding feasibility, acceptability, and signal for efficacy in the context of a clinical trial, combined with a dearth of evidence-based interventions to address loneliness in caregivers, future directions include study designs to facilitate rapid progression along the NIH Stage Model towards confirmation of efficacy (randomized trial with community interventionists; Stage III), additional study of biopsychosocial mechanisms to support dissemination strategies targeting active ingredients, and strategies to optimize delivery of Engage Coaching by diverse agencies and professionals. Subsequent trials could directly examine loneliness as a mechanism to improve health and well-being, with downstream (mediating) effects on reductions in caregiving strain, improved mental and physical health, improved physical and cognitive functioning, and increased quality of life, as loneliness affects all of these health indicators.